locale.c

它通过提供glibc兼容使得应用程序移植到较小的c 库时相当得容易. 它能够应用到带虚拟存储的Linux和uClinux上.在大多数带MMU部件的平台上为使它更加紧凑,它也能够编译成共享库.uClib

	uint16_t ii_len;
	uint16_t ti_len;
	uint16_t max_weight;
	uint16_t num_col_base;
	uint16_t max_col_index;
	uint16_t undefined_idx;
	uint16_t range_low;
	uint16_t range_count;
	uint16_t range_base_weight;
	uint16_t range_rule_offset;

	uint16_t index2weight_offset;
	uint16_t index2ruleidx_offset;
	uint16_t multistart_offset;
	uint16_t wcs2colidt_offset_low;
	uint16_t wcs2colidt_offset_hi;
} coldata_base_t;

typedef struct {
	uint16_t base_idx;
	uint16_t undefined_idx;
	uint16_t overrides_offset;
	uint16_t multistart_offset;
} coldata_der_t;

static int init_cur_collate(int der_num)
{
	__collate_t *cur_collate = &__global_locale.collate;
	const uint16_t *__locale_collate_tbl = __locale_mmap->collate_data;
	coldata_header_t *cdh;
	coldata_base_t *cdb;
	coldata_der_t *cdd;
	const uint16_t *p;
	size_t n;
	uint16_t i, w;

	assert(sizeof(coldata_base_t) == 19*2);
	assert(sizeof(coldata_der_t) == 4*2);
	assert(sizeof(coldata_header_t) == 8*2);

	if (!der_num) { 			/* C locale... special */
		cur_collate->num_weights = 0;
		return 1;
	}

	--der_num;

	cdh = (coldata_header_t *) __locale_collate_tbl;

	if (der_num >= cdh->num_der) {
		return 0;
	}

	cdd = (coldata_der_t *)(__locale_collate_tbl
							+ (sizeof(coldata_header_t)
							   + cdh->num_base * sizeof(coldata_base_t)
							   + der_num * sizeof(coldata_der_t)
							   )/2 );

	cdb = (coldata_base_t *)(__locale_collate_tbl
							 + (sizeof(coldata_header_t)
								+ cdd->base_idx * sizeof(coldata_base_t)
								)/2 );

	memcpy(cur_collate, cdb, offsetof(coldata_base_t,index2weight_offset));
	cur_collate->undefined_idx = cdd->undefined_idx;

	cur_collate->ti_mask = (1 << cur_collate->ti_shift)-1;
	cur_collate->ii_mask = (1 << cur_collate->ii_shift)-1;

/*	 printf("base=%d  num_col_base: %d  %d\n", cdd->base_idx ,cur_collate->num_col_base, cdb->num_col_base); */

	n = (sizeof(coldata_header_t) + cdh->num_base * sizeof(coldata_base_t)
		 + cdh->num_der * sizeof(coldata_der_t))/2;

/*	 printf("n   = %d\n", n); */
	cur_collate->index2weight_tbl = __locale_collate_tbl + n + cdb->index2weight_offset;
/*	 printf("i2w = %d\n", n + cdb->index2weight_offset); */
	n += cdh->num_index2weight;
	cur_collate->index2ruleidx_tbl = __locale_collate_tbl + n + cdb->index2ruleidx_offset;
/*	 printf("i2r = %d\n", n + cdb->index2ruleidx_offset); */
	n += cdh->num_index2ruleidx;
	cur_collate->multistart_tbl = __locale_collate_tbl + n + cdd->multistart_offset;
/*	 printf("mts = %d\n", n + cdb->multistart_offset); */
	n += cdh->num_multistart;
	cur_collate->overrides_tbl = __locale_collate_tbl + n + cdd->overrides_offset;
/*	 printf("ovr = %d\n", n + cdd->overrides_offset); */
	n += cdh->num_override;
	cur_collate->ruletable = __locale_collate_tbl + n;
/*	 printf("rtb = %d\n", n); */
	n += cdh->num_ruletable;
	cur_collate->weightstr = __locale_collate_tbl + n;
/*	 printf("wts = %d\n", n); */
	n += cdh->num_weightstr;
	cur_collate->wcs2colidt_tbl = __locale_collate_tbl + n
		+ (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16)
		+ cdb->wcs2colidt_offset_low;
/*	 printf("wcs = %lu\n", n	+ (((unsigned long)(cdb->wcs2colidt_offset_hi)) << 16) */
/* 		   + cdb->wcs2colidt_offset_low); */

	cur_collate->MAX_WEIGHTS = cdh->MAX_WEIGHTS;

#ifdef __UCLIBC_MJN3_ONLY__
#warning if calloc fails, this is WRONG.  there is also a memory leak here at the moment
#warning fix the +1 by increasing max_col_index?
#endif
	cur_collate->index2weight = calloc(2*cur_collate->max_col_index+2, sizeof(uint16_t));
	if (!cur_collate->index2weight) {
		return 0;
	}
	cur_collate->index2ruleidx = cur_collate->index2weight + cur_collate->max_col_index + 1;

	memcpy(cur_collate->index2weight, cur_collate->index2weight_tbl,
		   cur_collate->num_col_base * sizeof(uint16_t));
	memcpy(cur_collate->index2ruleidx, cur_collate->index2ruleidx_tbl,
		   cur_collate->num_col_base * sizeof(uint16_t));

	/* now do the overrides */
	p = cur_collate->overrides_tbl;
	while (*p > 1) {
/* 		fprintf(stderr, "processing override -- count = %d\n", *p); */
		n = *p++;
		w = *p++;
		do {
			i = *p++;
/* 			fprintf(stderr, "	i=%d w=%d *p=%d\n", i, w, *p); */
			cur_collate->index2weight[i-1] = w++;
			cur_collate->index2ruleidx[i-1] = *p++;
		} while (--n);
	}
	while (*++p) {
		i = *p;
		cur_collate->index2weight[i-1] = *++p;
		cur_collate->index2ruleidx[i-1] = *++p;
	}


	for (i=0 ; i < cur_collate->multistart_tbl[0] ; i++) {
		p = cur_collate->multistart_tbl;
/* 		fprintf(stderr, "%2d of %2d: %d ", i,  cur_collate->multistart_tbl[0], p[i]); */
		p += p[i];

		do {
			n = *p++;
			do {
				if (!*p) {		/* found it */
/* 					fprintf(stderr, "found: n=%d (%#lx) |%.*ls|\n", n, (int) *cs->s, n, cs->s); */
/* 					fprintf(stderr, ": %d - single\n", n); */
					goto FOUND;
 				}
				/* the lookup check here is safe since we're assured that *p is a valid colidex */
/* 				fprintf(stderr, "lookup(%lc)==%d  *p==%d\n", cs->s[n], lookup(cs->s[n]), (int) *p); */
/* 				fprintf(stderr, ": %d - ", n); */
				do {
/* 					fprintf(stderr, "%d|",  *p); */
				} while (*p++);
				break;
			} while (1);
		} while (1);
	FOUND:
		continue;
	}

	return 1;
}

void _locale_init(void)
{
	/* TODO: mmap the locale file  */

	/* TODO - ??? */
	memset(CUR_LOCALE_SPEC, 0, LOCALE_STRING_SIZE);
	CUR_LOCALE_SPEC[0] = '#';

	memcpy(__global_locale.category_item_count,
		   __locale_mmap->lc_common_item_offsets_LEN,
		   LC_ALL);

	++__global_locale.category_item_count[0]; /* Increment for codeset entry. */
	__global_locale.category_offsets[0] = offsetof(__locale_t, outdigit0_mb);
	__global_locale.category_offsets[1] = offsetof(__locale_t, decimal_point);
	__global_locale.category_offsets[2] = offsetof(__locale_t, int_curr_symbol);
	__global_locale.category_offsets[3] = offsetof(__locale_t, abday_1);
/*  	__global_locale.category_offsets[4] = offsetof(__locale_t, collate???); */
	__global_locale.category_offsets[5] = offsetof(__locale_t, yesexpr);

#ifdef __CTYPE_HAS_8_BIT_LOCALES
	__global_locale.tbl8ctype
		= (const unsigned char *) &__locale_mmap->tbl8ctype;
    __global_locale.tbl8uplow
		= (const unsigned char *) &__locale_mmap->tbl8uplow;
#ifdef __WCHAR_ENABLED
	__global_locale.tbl8c2wc
		= (const uint16_t *) &__locale_mmap->tbl8c2wc;
	__global_locale.tbl8wc2c
		= (const unsigned char *) &__locale_mmap->tbl8wc2c;
	/* translit  */
#endif /* __WCHAR_ENABLED */
#endif /* __CTYPE_HAS_8_BIT_LOCALES */
#ifdef __WCHAR_ENABLED
	__global_locale.tblwctype
		= (const unsigned char *) &__locale_mmap->tblwctype;
	__global_locale.tblwuplow
		= (const unsigned char *) &__locale_mmap->tblwuplow;
	__global_locale.tblwuplow_diff
		= (const uint16_t *) &__locale_mmap->tblwuplow_diff;
/* 	__global_locale.tblwcomb */
/* 		= (const unsigned char *) &__locale_mmap->tblwcomb; */
	/* width?? */
#endif /* __WCHAR_ENABLED */

	_locale_set(C_LOCALE_SELECTOR);
}

static const char ascii[] = "ASCII";
static const char utf8[] = "UTF-8";

void _locale_set(const unsigned char *p)
{
	const char **x;
	unsigned char *s = CUR_LOCALE_SPEC + 1;
	const size_t *stp;
	const unsigned char *r;
	const uint16_t *io;
	const uint16_t *ii;
	const unsigned char *d;
	int row;					/* locale row */
	int crow;					/* category row */
	int len;
	int c;
	int i = 0;

	++p;
	do {
		if ((*p != *s) || (p[1] != s[1])) {
			row  = (((int)(*p & 0x7f)) << 7) + (p[1] & 0x7f);
			assert(row < NUM_LOCALES);

			*s = *p;
			s[1] = p[1];

			if ((i != LC_COLLATE)
				&& ((len = __locale_mmap->lc_common_item_offsets_LEN[i]) != 0)
				) {
				crow = __locale_mmap->locales[ WIDTH_LOCALES * row + 3 + i ]
					* len;
				x = (const char **)(((char *) &__global_locale)
									+ __global_locale.category_offsets[i]);
				stp = __locale_mmap->lc_common_tbl_offsets + 4*i;
				r = (const unsigned char *)( ((char *)__locale_mmap) + *stp );
				io = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
				ii = (const uint16_t *)( ((char *)__locale_mmap) + *++stp );
				d = (const unsigned char *)( ((char *)__locale_mmap) + *++stp );
				for (c=0 ; c < len ; c++) {
					*(x + c) = d + ii[ r[crow + c] + io[c] ];
				}
			}
			if (i == LC_CTYPE) {
				c = __locale_mmap->locales[ WIDTH_LOCALES * row + 2 ]; /* codeset */
				if (c <= 2) {
					if (c == 2) {
						__global_locale.codeset = utf8;
						__global_locale.encoding = __ctype_encoding_utf8;
						/* TODO - fix for bcc */
						__global_locale.mb_cur_max = 6;
					} else {
						assert(c==1);
						__global_locale.codeset = ascii;
						__global_locale.encoding = __ctype_encoding_7_bit;
						__global_locale.mb_cur_max = 1;
					}
				} else {
					const codeset_8_bit_t *c8b;
					r = CODESET_LIST;
					__global_locale.codeset = r + r[c -= 3];
					__global_locale.encoding = __ctype_encoding_8_bit;
#ifdef __UCLIBC_MJN3_ONLY__
#warning REMINDER: update 8 bit mb_cur_max when trasnlit implemented!
#endif
					/* TODO - update when translit implemented! */
					__global_locale.mb_cur_max = 1;
					c8b = __locale_mmap->codeset_8_bit + c;
#ifdef __CTYPE_HAS_8_BIT_LOCALES
					__global_locale.idx8ctype = c8b->idx8ctype;
					__global_locale.idx8uplow = c8b->idx8uplow;
#ifdef __WCHAR_ENABLED
					__global_locale.idx8c2wc = c8b->idx8c2wc;
					__global_locale.idx8wc2c = c8b->idx8wc2c;
					/* translit  */
#endif /* __WCHAR_ENABLED */
#endif /* __CTYPE_HAS_8_BIT_LOCALES */
				}
#ifdef __UCLIBC_MJN3_ONLY__
#warning might want to just put this in the locale_mmap object
#endif
				d = __global_locale.outdigit_length;
				x = &__global_locale.outdigit0_mb;
				for (c = 0 ; c < 10 ; c++) {
					((unsigned char *)d)[c] = strlen(x[c]);
					assert(d[c] > 0);
				}
			} else if (i == LC_COLLATE) {
				init_cur_collate(__locale_mmap->locales[ WIDTH_LOCALES * row + 3 + i ]);
			}
		}
		++i;
		p += 2;
		s += 2;
	} while (i < LC_ALL);
}

#endif /* __LOCALE_C_ONLY */

#endif
/**********************************************************************/
#ifdef L_nl_langinfo

#include <langinfo.h>
#include <nl_types.h>

#ifdef __LOCALE_C_ONLY

/* We need to index 320 bytes of data, so you might initially think we
 * need to store the offsets in shorts.  But since the offset of the
 * 64th item is 182, we'll store "offset - 2*64" for all items >= 64
 * and always calculate the data offset as "offset[i] + 2*(i & 64)".
 * This allows us to pack the data offsets in an unsigned char while
 * also avoiding an "if".
 *
 * Note: Category order is assumed to be:
 *   ctype, numeric, monetary, time, collate, messages, all
 */

#define C_LC_ALL 6

/* Combine the data to avoid size penalty for seperate char arrays when
 * compiler aligns objects.  The original code is left in as documentation. */
#define cat_start nl_data
#define C_locale_data (nl_data + C_LC_ALL + 1 + 90)

static const unsigned char nl_data[C_LC_ALL + 1 + 90 + 320] = {
/* static const char cat_start[LC_ALL + 1] = { */
        '\x00', '\x0b', '\x0e', '\x24', '\x56', '\x56', '\x5a', 
/* }; */
/* static const char item_offset[90] = { */
	'\x00', '\x02', '\x04', '\x06', '\x08', '\x0a', '\x0c', '\x0e', 
	'\x10', '\x12', '\x14', '\x1a', '\x1b', '\x1b', '\x1b', '\x1b', 
	'\x1b', '\x1b', '\x1b', '\x1b', '\x1b', '\x1c', '\x1c', '\x1c', 
	'\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', '\x1c', 
	'\x1c', '\x1c', '\x1c', '\x1e', '\x20', '\x24', '\x28', '\x2c', 
	'\x30', '\x34', '\x38', '\x3c', '\x43', '\x4a', '\x52', '\x5c', 
	'\x65', '\x6c', '\x75', '\x79', '\x7d', '\x81', '\x85', '\x89', 
	'\x8d', '\x91', '\x95', '\x99', '\x9d', '\xa1', '\xa5', '\xad', 
	'\x36', '\x3c', '\x42', '\x46', '\x4b', '\x50', '\x57', '\x61', 
	'\x69', '\x72', '\x7b', '\x7e', '\x81', '\x96', '\x9f', '\xa8', 
	'\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb3', '\xb4', '\xba', 
	'\xbf', '\xbf', 
/* }; */
/* static const char C_locale_data[320] = { */
	   '0', '\x00',    '1', '\x00',    '2', '\x00',    '3', '\x00', 
	   '4', '\x00',    '5', '\x00',    '6', '\x00',    '7', '\x00', 
	   '8', '\x00',    '9', '\x00',    'A',    'S',    'C',    'I', 
	   'I', '\x00',    '.', '\x00', '\x7f', '\x00',    '-', '\x00', 
	   'S',    'u',    'n', '\x00',    'M',    'o',    'n', '\x00', 
	   'T',    'u',    'e', '\x00',    'W',    'e',    'd', '\x00', 
	   'T',    'h',    'u', '\x00',    'F',    'r',    'i', '\x00', 
	   'S',    'a',    't', '\x00',    'S',    'u',    'n',    'd', 
	   'a',    'y', '\x00',    'M',    'o',    'n',    'd',    'a', 
	   'y', '\x00',    'T',    'u',    'e',    's',    'd',    'a', 
	   'y', '\x00',    'W',    'e',    'd',    'n',    'e',    's', 
	   'd',    'a',    'y', '\x00',    'T',    'h',    'u',    'r', 
	   's',    'd',    'a',    'y', '\x00',    'F',    'r',    'i', 
	   'd',    'a',    'y', '\x00',    'S',    'a',    't',    'u', 
	   'r',    'd',    'a',    'y', '\x00',    'J',    'a',    'n', 
	'\x00',    'F',    'e',    'b', '\x00',    'M',    'a',    'r', 
	'\x00',    'A',    'p',    'r', '\x00',    'M',    'a',    'y', 
	'\x00',    'J',    'u',    'n', '\x00',    'J',    'u',    'l', 
	'\x00',    'A',    'u',    'g', '\x00',    'S',    'e',    'p', 
	'\x00',    'O',    'c',    't', '\x00',    'N',    'o',    'v', 
	'\x00',    'D',    'e',    'c', '\x00',    'J',    'a',    'n', 
	   'u',    'a',    'r',    'y', '\x00',    'F',    'e',    'b', 
	   'r',    'u',    'a',    'r',    'y', '\x00',    'M',    'a', 
	   'r',    'c',    'h', '\x00',    'A',    'p',    'r',    'i', 
	   'l', '\x00',    'M',    'a',    'y', '\x00',    'J',    'u', 
	   'n',    'e', '\x00',    'J',    'u',    'l',    'y', '\x00', 
	   'A',    'u',    'g',    'u',    's',    't', '\x00',    'S', 
	   'e',    'p',    't',    'e',    'm',    'b',    'e',    'r', 
	'\x00',    'O',    'c',    't',    'o',    'b',    'e',    'r', 
	'\x00',    'N',    'o',    'v',    'e',    'm',    'b',    'e', 
	   'r', '\x00',    'D',    'e',    'c',    'e',    'm',    'b', 
	   'e',    'r', '\x00',    'A',    'M', '\x00',    'P',    'M', 
	'\x00',    '%',    'a',    ' ',    '%',    'b',    ' ',    '%', 
	   'e',    ' ',    '%',    'H',    ':',    '%',    'M',    ':', 
	   '%',    'S',    ' ',    '%',    'Y', '\x00',    '%',    'm', 
	   '/',    '%',    'd',    '/',    '%',    'y', '\x00',    '%', 
	   'H',    ':',    '%',    'M',    ':',    '%',    'S', '\x00', 
	   '%',    'I',    ':',    '%',    'M',    ':',    '%',    'S', 
	   ' ',    '%',    'p', '\x00',    '^',    '[',    'y',    'Y', 
	   ']', '\x00',    '^',    '[',    'n',    'N',    ']', '\x00', 
};

char *nl_langinfo(nl_item item)
{
	unsigned int c;
	unsigned int i;

	if ((c = _NL_ITEM_CATEGORY(item)) < C_LC_ALL) {
		if ((i = cat_start[c] + _NL_ITEM_INDEX(item)) < cat_start[c+1]) {
/*  			return (char *) C_locale_data + item_offset[i] + (i & 64); */
			return (char *) C_locale_data + nl_data[C_LC_ALL+1+i] + 2*(i & 64);
		}
	}
	return (char *) cat_start;	/* Conveniently, this is the empty string. */
}

#else  /* __LOCALE_C_ONLY */

static const char empty[] = "";

char *nl_langinfo(nl_item item)
{
	unsigned int c = _NL_ITEM_CATEGORY(item);
	unsigned int i = _NL_ITEM_INDEX(item);

	if ((c < LC_ALL) && (i < __global_locale.category_item_count[c])) {
		return ((char **)(((char *) &__global_locale)
						  + __global_locale.category_offsets[c]))[i];

	}
	return (char *) empty;
}

#endif /* __LOCALE_C_ONLY */

#endif
/**********************************************************************/